Again so close to agreeing
@Cosmik!
The only reason I can’t fully agree is because I reject that it’s logically possible to have an even theoretically “perfect” speaker.
There’s just no information in the recording about the polar response of the reproduction system (nor the room for that matter).
To state my point in different terms: the recording is 2D (amplitude and frequency only) while the reproduction takes place in 3 dimensions.
So when it comes to this question, there is simply nothing in the recording in the first place for the reproduction to try to be true to.
By definition there is no “perfect” polar response.
Ofc constant directivity or a smoothly downward sloping power response are
IMO the best things to aim for usually, but even choices between these or between versions of these come down partly to preference, partly to room, and partly to objectives (eg mixing vs mastering vs listening vs dancing etc).
There’s just no absolute, definitive way of solving this one...
EDIT: except perhaps by agreed-upon and widely adopted convention?